Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752928Ab1BRFyC (ORCPT ); Fri, 18 Feb 2011 00:54:02 -0500 Received: from sj-iport-3.cisco.com ([171.71.176.72]:29254 "EHLO sj-iport-3.cisco.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752299Ab1BRFx6 (ORCPT ); Fri, 18 Feb 2011 00:53:58 -0500 Authentication-Results: sj-iport-3.cisco.com; dkim=neutral (message not signed) header.i=none X-IronPort-AV: E=Sophos;i="4.62,185,1297036800"; d="scan'208";a="265626818" From: David Ahern To: linux-perf-users@vger.kernel.org, linux-kernel@vger.kernel.org Cc: peterz@infradead.org, mingo@elte.hu, acme@ghostprotocols.net, paulus@samba.org, David Ahern Subject: [PATCH 2/3] perf events: Introduce realtime clock event Date: Thu, 17 Feb 2011 22:53:52 -0700 Message-Id: <1298008433-22911-3-git-send-email-daahern@cisco.com> X-Mailer: git-send-email 1.7.3.4 In-Reply-To: <1298008433-22911-1-git-send-email-daahern@cisco.com> References: <1298008433-22911-1-git-send-email-daahern@cisco.com> Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 4404 Lines: 149 The motivation for this event is to convert perf_clock() time stamps to wall-clock (gettimeofday()) equivalents, including adjustments made by NTP (e.g., for comparing perf events to other log files). This patch is based on the monotonic patch by Arnaldo Carvalho de Melo . (NOTE: Comments from the last review of the timehist patch series suggested calling this a monotonic clock. I am not trying to be dense here; since gettimeofday maps to realtime clock I think that is the better name for it.) Signed-off-by: David Ahern --- include/linux/perf_event.h | 1 + kernel/perf_event.c | 72 ++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 73 insertions(+), 0 deletions(-) diff --git a/include/linux/perf_event.h b/include/linux/perf_event.h index 8ceb5a6..51a2f34 100644 --- a/include/linux/perf_event.h +++ b/include/linux/perf_event.h @@ -105,6 +105,7 @@ enum perf_sw_ids { PERF_COUNT_SW_PAGE_FAULTS_MAJ = 6, PERF_COUNT_SW_ALIGNMENT_FAULTS = 7, PERF_COUNT_SW_EMULATION_FAULTS = 8, + PERF_COUNT_SW_REALTIME_CLOCK = 9, PERF_COUNT_SW_MAX, /* non-ABI */ }; diff --git a/kernel/perf_event.c b/kernel/perf_event.c index 3280671..4b6d29c 100644 --- a/kernel/perf_event.c +++ b/kernel/perf_event.c @@ -5413,6 +5413,7 @@ static int perf_swevent_init(struct perf_event *event) switch (event_id) { case PERF_COUNT_SW_CPU_CLOCK: case PERF_COUNT_SW_TASK_CLOCK: + case PERF_COUNT_SW_REALTIME_CLOCK: return -ENOENT; default: @@ -5599,6 +5600,7 @@ static enum hrtimer_restart perf_swevent_hrtimer(struct hrtimer *hrtimer) struct perf_sample_data data; struct pt_regs *regs; struct perf_event *event; + struct perf_raw_record raw; u64 period; event = container_of(hrtimer, struct perf_event, hw.hrtimer); @@ -5610,6 +5612,13 @@ static enum hrtimer_restart perf_swevent_hrtimer(struct hrtimer *hrtimer) perf_sample_data_init(&data, 0); data.period = event->hw.last_period; + if (event->attr.sample_type & PERF_SAMPLE_RAW) + { + raw.size = sizeof(u64); + raw.data = &event->count; + data.raw = &raw; + } + regs = get_irq_regs(); if (regs && !perf_exclude_event(event, regs)) { @@ -5751,6 +5760,68 @@ static struct pmu perf_cpu_clock = { }; /* + * Software event: realtime wall time clock + */ + +static void realtime_clock_event_update(struct perf_event *event) +{ + u64 now = ktime_to_ns(ktime_get_real()); + local64_set(&event->count, now); +} + +static void realtime_clock_event_start(struct perf_event *event, int flags) +{ + realtime_clock_event_update(event); + perf_swevent_start_hrtimer(event); +} + +static void realtime_clock_event_stop(struct perf_event *event, int flags) +{ + perf_swevent_cancel_hrtimer(event); + realtime_clock_event_update(event); +} + +static int realtime_clock_event_add(struct perf_event *event, int flags) +{ + if (flags & PERF_EF_START) + realtime_clock_event_start(event, flags); + + return 0; +} + +static void realtime_clock_event_del(struct perf_event *event, int flags) +{ + realtime_clock_event_stop(event, flags); +} + +static void realtime_clock_event_read(struct perf_event *event) +{ + realtime_clock_event_update(event); +} + +static int realtime_clock_event_init(struct perf_event *event) +{ + if (event->attr.type != PERF_TYPE_SOFTWARE) + return -ENOENT; + + if (event->attr.config != PERF_COUNT_SW_REALTIME_CLOCK) + return -ENOENT; + + return 0; +} + +static struct pmu perf_realtime_clock = { + .task_ctx_nr = perf_sw_context, + + .event_init = realtime_clock_event_init, + .add = realtime_clock_event_add, + .del = realtime_clock_event_del, + .start = realtime_clock_event_start, + .stop = realtime_clock_event_stop, + .read = realtime_clock_event_read, +}; + +/* * Software event: task time clock */ @@ -7285,6 +7356,7 @@ void __init perf_event_init(void) init_srcu_struct(&pmus_srcu); perf_pmu_register(&perf_swevent, "software", PERF_TYPE_SOFTWARE); perf_pmu_register(&perf_cpu_clock, NULL, -1); + perf_pmu_register(&perf_realtime_clock, NULL, -1); perf_pmu_register(&perf_task_clock, NULL, -1); perf_tp_register(); perf_cpu_notifier(perf_cpu_notify); -- 1.7.3.4 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/